The bowels of the Soufrière volcano in Guadeloupe mapped in 3D

Overlooked by the Soufrière volcano, the highest point of Guadeloupe, the inhabitants of the West Indian island are among the 800 million people in the world who live less than 100 kilometers from a dangerous volcano, and are therefore subject to the whims of of the one they nickname “the Old Lady”. To best manage the risk linked to volcanic eruptions, it is essential to predict their occurrence, which is often very difficult due to a lack of data on the interior of the volcano. Elsa Giraudat, of the Langevin Institute, in Paris, and her colleagues at the Institute of Globe Physics have made an important breakthrough by producing three-dimensional images of the internal structure of Soufrière.

The team installed a network of 76 geophones on the sides of the volcano, devices that measure seismic noise, caused in particular by human activities or waves on the sand. When seismic waves encounter reflective structures in the volcano (fractures, magma pockets, etc.), they change trajectory. It would in principle be possible, by measuring them on the surface, to reconstruct the internal structure of the building. Alexandre Aubry, research director at the CNRS, who participated in the study, explains: “For a long time, we did not use seismic noise, even though it contains a lot of information. Researchers realized around twenty years ago that if we correlate the seismic noise recorded by two geophones, we access what we call the “impulse response” between these geophones: everything happens as if we were creating a microseism at one of the devices and that we listened to it from the other. By considering all possible combinations of pairs, we obtain a reflection matrix, from which we construct an image of the reflective structures of the volcano, as in ultrasound. »

However, the image thus collected is blurred by the heterogeneities of the environment, which create distortions. To correct these aberrations, the researchers applied computer processing to their data, drawing inspiration from a method used in astronomy with light waves. “When we look at a star,” continues Alexandre Aubry, “we see it sparkle because of the turbulence in the atmosphere which distorts its image. Astronomers compensate for these aberrations using deformable mirrors placed in the telescope. » In this technique called “adaptive optics”, astronomers use a laser pointed towards the sky which excites the sodium atoms present in the upper atmosphere and thus creates a “virtual star”. The image distorted by the turbulence of this artificial star indicates how to configure the mirror to collect a clear image of real stars. “Here, we take up this principle,” continues Alexandre Aubry. From the data from the reflection matrix, we synthesize kinds of virtual stars in all areas of the volcano and we “listen” to the seismic waves coming from these “stars” to study and compensate for their distortion. »

The researchers thus obtained a second, much sharper image of the interior of the volcano, with a precision of around a hundred meters. They discovered that between 5 kilometers depth and the surface of Soufrière, the magma crosses the rock in a long tortuous conduit, while between 8 and 5 kilometers depth, it is distributed in large horizontal lenses superimposed on each other. others and connected by small vertical conduits.

Thanks to this unique method, Alexandre Aubry and his colleagues hope to go further: “The next step would be to establish this kind of image as a function of time, to study the dynamics of the magma in depth. Furthermore, currently, to predict eruptions, we listen to microseisms at depth without really knowing where they come from. Our method would therefore make it possible to better locate these earthquakes to know which underground zone of the volcano is moving. » These data would also help to accurately assess the pressures and temperatures within the volcano, the connectivity between magmatic systems and the surface, to detect the presence of potentially eruptible magma earlier and to monitor its spatiotemporal evolution… parameters that influence the occurrence of volcanic eruptions.